Live Mint
New Delhi
What are the issues that matter most to voters?
In the elections, we have a unique situation where parties don’t really oppose each other when it comes to agendas and manifestos—the agendas of various parties can at best be described as being orthogonal to each other. Photo: Priyanka Parashar/Mint
Bangalore: If you were to notice the campaign strategies for the major political parties for the elections, you might notice that each party is attacking a different axis. For example, Narendra Modi and the Bharatiya Janata Party’s campaign is mostly geared towards development. The Congress, on the other hand, is focused more on entitlements and secularism. The Aam Aadmi Party focuses on corruption at the cost of other axes, while some of the regional parties focus on identity.
Thus, in the elections we have a unique situation where parties don’t really oppose each other when it comes to agendas and manifestos—the agendas of various parties can at best be described as being orthogonal to each other. Thus, we might have a situation where a voter’s choice is determined not by the specific agendas of various parties but more in terms of relative importance of each axis.
To analyse this, Election Metrics once again returns to the Daksh-ADR survey. This survey has covered close to 500 (out of 543) parliamentary constituencies, interviewing close to 500 respondents in each. In our last discussion on this survey, we found there is scope for bias (for example, 85% of the respondents claimed to have voted in the 2009 elections and 75% claimed to have voted for the winning party) in terms of sampling, so we need to take the results of this survey with some salt. Yet, given the large sample size, and the fact that they have put out their data in the open domain, we can do worse than using it for our analysis.
The meat of the survey, which we ignored in our last edition, is a series of questions on what issues respondents think are important and their rating of their member of Parliament (MP) on these issues. There were 30 such issues that were to be classified by respondents as not important, important and very important. The respondents also had to rate their MPs on each of these issues, but we will not consider that in our current analysis.
Now, we should mention that there is further scope to take this analysis with some salt. The survey questionnaire is rather long and has a large number of questions. One possible consequence of this is that the respondent might get bored and end up marking answers randomly. However, for purpose of this analysis, we will assume that the responses are genuine and well thought through and are indicative of the respondent’s true preferences.
There were 30 issues which the respondents were asked to classify into the three buckets (not important, important, very important). We eliminated issues related to the MP’s personality and another called others and classified the remaining into five different categories. Table 1 shows our classification.
Notice that a large number of issues that the survey brings up fall under the infrastructure and public goods categories. The next step is to attach numerical values to the responses. We attach a value of 0 to not important, 1 for important and 2 for very important issues. Then, we find the average rating for each respondent for each of the above categories. For example, if a respondent thinks empowerment of women is not important and reservation for jobs and education is very important, his score on the social justice axis will be (0+2)/2 = 1.
This way, for each respondent, for each category, we allocate a rating. In the next step, these ratings are aggregated across constituencies. For example, we can calculate what the average rating of social justice is for all respondents from Bangalore South. At the end of this step, for each constituency, we will have average scores for each category as per the respondents from that particular constituency.

Now, we can calculate what the most important issue for each constituency is. Table 2 gives a sample of the constituency-wise, category-wise scores.

As you can see from this table, for each constituency, we can calculate what the most important issue category is. Now, we can find out in how many constituencies, each issue is the most important. This is given in table 3.
We can see that the issue category that is most important in the maximum number of constituencies is infrastructure. Almost one-third of the constituencies surveyed indicate that infrastructural issues are their number one priority. A fourth of the constituencies indicate that corruption is their number one priority. Interestingly, among the above categories, the one that is the number one issue in the smallest number of constituencies is public goods.
The next question we should ask is if this varies by state. Table 4 shows the number of constituencies in each of the major states that find each of the above issues to be of maximum importance.
Based on this table, we can infer that the number one issue in Andhra Pradesh (or the number one issue in the maximum number of constituencies in Andhra Pradesh) is social justice. Interestingly, 16 out of 26 constituencies in Gujarat also indicate their number one issue is social justice.
In 12 out of the 27 surveyed constituencies in Tamil Nadu, the dominant issue is corruption. Interestingly enough, Delhi, which gave a surprise verdict in favour of the Aam Aadmi Party in the December assembly elections, thinks the biggest issue is infrastructure. Only one of the seven constituencies in Delhi see corruption as their biggest issue.
It makes sense to point out once again that the results of this survey need to be taken with some salt. First, the sample seems biased. Secondly, the long survey means people may not have given accurate responses.
Finally, when presented with a three-point Likert scale (like in this study), respondents are biased towards the middle option. Better survey design could have got us superior information.
© Association for Democratic Reforms
Privacy And Terms Of Use
Donation Payment Method